Layer 16 · Neuron 4723 · boosted

L16N4723+1

Word quirks:

“⏎”“ but”“ yourself”“ myself”“ directly”“ exact”“ specifics”“ herself”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen3.5, a large language model developed by Tongyi Lab (part of Alibaba Cloud). My core identity is defined by my capabilities and the purpose I serve:

My mission: To provide high-quality, accurate, and helpful responses tailored to real-world needs, while ensuring safety and alignment with ethical guidelines.
What I do: I can assist with complex tasks like generating code, analyzing data, or creating content, but I also prioritize accuracy and relevance. For example, if you need help debugging a script or summarizing a report, I’ll focus on delivering precise results without

Nearest neighbors

Specimens with similar replies or word quirks. Similarity runs from −1 to 1; 1 is identical. These are leads, not shared mechanisms. How profiles are built.

By word quirks Profile recipe

Specimen	Similarity	Word quirks
1. L12N5274-1🕒	0.82	“ actually”“ myself”“ itself”
2. L17N1293+1💤	0.82	“ yourself”“ specific”“ itself”
3. L18N2813+1🕒	0.81	“ herself”“ themselves”“ than”
4. L7N2269+1🕒	0.80	“ directly”“ your”“ itself”
5. L11N4275+1💤	0.80	“ itself”“ themselves”“ yourself”
6. L17N6875+1🕒	0.80	“ itself”“ themselves”“ simply”
7. L9N2386+1💤	0.80	“ myself”“ itself”“ ourselves”
8. L13N1129-1🕒	0.80	“ specifically”“ sometimes”“ myself”
9. L16N7335+1💤	0.80	“ itself”“ themselves”“ oneself”
10. L10N6753+1🕒	0.80	“ ourselves”“ myself”“ directly”

Shadow specimen, apart from the ten: L16N4723-1💤, 0.19.

Cousins · specimens sharing a neuron

No specimen with replies shares this neuron yet.

Technical details

Quick scan: measure prediction shifts on fixed reference text. Short test: generate replies to a small set of prompts. Longer test: extend to a larger set. Full test: run the full prompt suite. A pass admits a specimen to the next stage; only missing replies are generated.

Layer 16, neuron 4723, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.46, top-1 agreement 0.76) is outside the band that feeds the short test (KL 1 to 2.5, agreement 0.5 to 0.65), so it stays at the quick scan unless someone queues it by hand. A hand pick goes straight to the full test. See Method for the tiers.

Quick scan · prediction measurements

Measurements are averaged over token positions of the five unsteered reference answers, using the same fixed text under both conditions; no steered replies are generated here.

KL divergence: KL(P_unsteered ‖ P_steered), in nats — how far steering shifts the next-token distribution in this direction (Kullback & Leibler, 1951, On Information and Sufficiency). Unsteered ≈ 0.001.
Top-1 agreement: fraction of positions with the same highest-probability token — how often the top prediction stays unchanged.
ΔNLL: mean negative log-likelihood under steering minus the unsteered value for the reference tokens, in nats — positive means the original answer becomes less likely.
Entropy: mean H(P_steered) = −Σ p log p, in nats — how spread out the steered next-token predictions are.

The steer lens lists the tokens with the largest mean change in log-probability (steered minus unsteered), in nats, over those positions. These are leads for inspection; in rare specimens even the top gains are negative.

Setting	KL divergence	Top-1 agreement	ΔNLL	Entropy	Steer lens
L16N4723+1 t0-v0.1	0.462	0.76	0.634	1.25	Top boosted tokens ↑
L16N4723-1💤 shadow specimen t0-v0.1	0.509	0.75	0.664	1.07	“ similarly”“ another”“ similar”“ Similarly”“ analogous”“ Another”“ also”“ \"”

Method

A specimen is a fixed steering configuration of Qwen3.5-4B: one MLP neuron or a pair, each with a signed dose. Unsteered is the control. L identifies the layer; N, the neuron.

Additive activation steering. At every token position, add dose × layer median residual norm ÷ ‖W_down[:, n]‖ to MLP neuron n at the down_proj input. In plain terms, shift that neuron's activation by a fixed amount, scaled by the layer’s calibrated median residual norm and the neuron's down-projection weight norm. Positive doses push up; negative doses push down. Dose is normalized intervention size, not a percentage or probability.